Social network analysis is the study of the connections and relationships between individual entities, and is increasingly proving to be a powerful and elegant toolkit for studying complex systems arising from many scientific disciplines. In this talk, I discuss the topics of community detection and graph probing, with approaches including algorithm development, evaluation of existing techniques, and interdisciplinary application of social network analysis techniques.
Community detection and understanding community structure are critical to a variety of fields, including ecology, sociology, and medicine. As I discuss: Which mountain lion populations are at greatest risk of disease? How does the behavior of individual lions relate to that of their community? Despite the value of communities, there is little consensus within the research literature as to what a community actually is, and characterizing the structure of real communities is an important research challenge. I present a machine-learning framework to better understand the structure of real-world communities, as well as applications of community detection.
Active graph probing is a new and challenging social network analysis problem, also with applications in a variety of disciplines, including military intelligence and cybersecurity. This problem considers a user interacting with a partially-observed graph. Given a budget that can be used to probe for more information about selected nodes in the network, which nodes should be selected in order to gain the most useful information about the true network? This problem presents major algorithmic challenges, and I present our initial approaches towards solving it.
Dr. Arun Ross